Multi-level human-machine-interaction monitoring and system reconfiguration

Abstract

A hierarchical approach for human-machine-interaction monitoring and system reconfiguration is proposed, which is grounded on local quality of interaction detection and mode-switching in real-time. For integrating the human behaviour in the closed-loop control structure, the overall technical system is decomposed recursively into nested behaviour levels and interfaced to the user on the corresponding level. In a gaze-based wheelchair control application, the local interaction with the user’s gaze is assessed in order to estimate the intention for a specific action related to wheelchair control. Besides the level of interest directed to a goal object, the quality of estimation is forwarded to a reconfiguration component. Finally, the system is switched to the interaction mode, which provides maximum assistance under the given interaction constraints.